Geometric calibration of all-sky cameras using sun and moon positions: A comprehensive analysis

Abstract

All-sky imagers (ASIs) have been applied to enable accurate very-short-term forecasts of the production of solar power plants and to derive measurements which facilitate the efficient and reliable operation of such plants. Overall, ASIs can support the grid integration and efficient operation of solar power plants.

These tasks often require to reconstruct ‘world coordinates’ of observed scene points from their position in the ASI images. This requires a practically feasible geometric calibration of each ASI regarding camera-intrinsic lens distortion parameters and the ASI’s external orientation.

We present ‘SuMo’ an open-source Python tool which determines all relevant parameters only using regular ASI images of Sun and Moon. The method avoids a manual interference on-site, can be applied retrospectively and can also be used to continuously monitor an ASI’s geometric calibration.

We validate the calibration method on five cameras at three sites and over various datasets representing different seasons, atmospheric conditions, exposure times and sun/ moon elevation and azimuth angles. Already a single month of images from either summer or winter yields an accurate calibration (RMSE $\le 0.14^{\circ }$). A comparable calibration accuracy (RMSE 0.14 – $0.38^{\circ }$) could be achieved for all tested ASIs without modifying any of the method’s parameters. Image quality moderately influenced the calibration accuracy. An additional cross-validation with the star-based ORION calibration method further confirms the high accuracy of our method over the entire sky dome (MAE $0.14^{\circ }$).

We provide a Python package and $>2$ years of ASI images and irradiance measurements with the publication.